Semilocal Features (SL)
Every packaged CIDER model begins with a semilocal block built from the density \(n(\mathbf r)=\sum_i f_i|\phi_i(\mathbf r)|^2\), its gradient, and for meta-GGA models the kinetic-energy density
Regularized combinations of these ingredients can encode physical constraints such as uniform scaling. They include the reduced gradient[1]
and the iso-orbital indicator[2]
where \(\tau_W=|\nabla n|^2/8n\) is the single-electron kinetic energy density, and \(\tau_0=\frac{3}{10}(3\pi^2)^{2/3}n^{5/3}\) is the kinetic energy density of the uniform electron gas.
These equations use the spin-unpolarized convention, with orbital occupations \(f_i\in [0,2]\).
In CiderPress, SemilocalSettings specifies
the semilocal block. Its mode selects one of four raw feature layouts:
ns: \(\mathbf{x}_\text{sl} = [n, |\nabla n|^2]\)np: \(\mathbf{x}_\text{sl} = [n, p]\)nst: \(\mathbf{x}_\text{sl} = [n, |\nabla n|^2, \tau]\)npa: \(\mathbf{x}_\text{sl} = [n, p, \alpha]\)
The ns and np layouts omit the kinetic-energy density \(\tau\).
An associated NLDF block therefore uses
sl_level="GGA". Orbital-dependent features such as SDMX may still be present. Modes nst and npa include
\(\tau\) and are meta-GGA level.
The regularized features \(p\) and \(\alpha\) are scale-invariant, meaning that an exchange functional trained with these features and a proper exchange functional baseline will obey the uniform scaling rule (see Uniform Scaling Constraints).
The raw features \(n\), \(|\nabla n|^2\), and \(\tau\) carry nonzero scaling powers. During model construction, normalizers and feature maps supply the behavior required by the chosen functional form. Those transformations are serialized with a trained model and are applied automatically when a packaged functional is evaluated.
See SemilocalSettings in the
Feature Settings
documentation for more details on the API for setting up these features.